Interactions between transcription factors (TFs) and their DNA binding sites are an integral part of regulatory networks within cells. These interactions control critical steps in progression through normal cellular processes and in responses to various environmental stresses. However, the DNA binding site specificities and regulatory functions of many known and most predicted TFs in S. cerevisiae are unknown. The Bulyk Lab has recently developed an improved in vitro protein binding microarray (PBM) technology to characterize TFs' sequence specificities in a high-throughput manner. The PBM technology allows us to determine the binding site specificity of known or predicted transcription factors in a single day, starting from the purified TF. Preliminary comparisons of binding site specificities determined from the PBM approach correspond well with binding site specificities determined by in vivo genome-wide location analysis. We propose to characterize the DNA binding site specificities of all known and predicted TFs whose binding specificities are as yet unknown. To achieve this goal, the Bulyk Lab has teamed up with Dr. Joshua Labaer's Lab (Harvard Medical School and the Harvard Institute of Proteomics), whose expertise with highthroughput cloning, expression, and purification will enable us to examine a collection of ~300 candidate yeast TFs, and with Dr. Richard Young (MIT / Whitehead Institute) and the Whitehead Institute's Microarray Center, who will be providing whole-genome yeast intergenic microarrays for use in PBM experiments. The TFs' binding sites will enable us to predict the TFs' functional roles by examining the functions of the candidate target genes. These candidate functional roles will be examined experimentally by assessing the phenotypes of the corresponding TF deletion strains under conditions in which the particular TFs are hypothesized to regulate the candidate target genes. Comparison of the in vitro PBM data and the in vivo genome-wide location analysis data may provide data as to why certain sites are or are not used in vivo. These studies will also serve as a model system for applying the PBM technology on a genome- and proteome-wide scale. In the future this kind of scaled-up PBM approach could be used to examine a large collection of candidate combinatorial TF interactions in DNA binding. This PBM approach could also be used to study TFs from other genomes. We will make our data publicly available in a web-accessible database.